New Requirements for Reference Safety Information in the EU from January 2019

The European Clinical Trials Facilitation Group published a Q&A document in 2017 addressing the requirements for Reference Safety Information (RSI) for medicinal products being used in clinical trials.  The RSI is used for expectedness assessment for adverse reactions reported in clinical trials for two purposes:

  1. To help identify SUSARs which are subject to expedited reporting
  2. To categorise events reported in the Development Safety Update Report (DSUR) which is sent annually to regulators and ethics committees

The requirements outlined in this document become mandatory from January 2019 and so it is important for those involved in any aspect of ADR reporting to be aware of them.

What should be included in the RSI.

batch books document education
Photo by Pixabay on

  1. There should be a specific section in the Investigators Brochure entitled “Reference safety information for assessment of expectedness of serious adverse reactions”and this section should be the basis for expectedness assessment, not the general safety information section. An event can only be considered expected if it appears at the same frequency, specificity and intensity in the RSI section in the IB. For licensed products the Summary of Product Characteristics (SmPC) can be used as the RSI if the product is used within the terms of its marketing authorisation and the SmPC is identified as the source of the RSI in the Clinical Trials Authorisation (CTA) application cover letter.
  2. The “RSI is a list of expected serious adverse reactions, which are classified using Preferred Terms (PTs) according to the Medical Dictionary for Regulatory Activities (MedDRA)”
  3. Expected events should only be included in the RSI if they are Serious and meet the following criteria:
    1. There is reasonable evidence of a causal relationship to the IMP after through review by the sponsor
    2. The event has been reported as a suspected serious adverse reaction more than once
    3. The event has been observed with this product. You can’t include events which might be anticipated from the pharmacology of the product or the product class.
  4. There’s a section on fatal and life-threatening events, which, being honest, I find a little confusing. However my reading of it is that generally speaking there should be no fatal or life-threatening SARs listed in the RSI. The argument is that sponsors should not expect a product to cause fatal or life-threatening reactions unless there is a robust risk-benefit analysis to support this.  This means that fatal and life-threatening SARs will usually be unexpected even if fatal or life-threatening events have been seen previously.  For a marketed product a fatal or life-threatening event can be considered expected if the event is listed in the SmPC, but for unlicensed products there should never be fatal or life-threatening events in the RSI.
  5. What about events where the causality assessment is unlikely or possible? The advice is as follows: “An adverse reaction, in contrast to an adverse event, is characterized by the fact that a causal relationship between a medicinal product and an occurrence is suspected. Thus in a clinical trial setting, a causal relationship to the IMP is either considered to be suspected or not for each individual adverse event which occurs. If an investigator uses the WHO classification categories of causality when assessing causality, ‘highly probable’, ‘probable’, ‘possible’ should be regarded as related by the sponsor, while ‘unlikely’ and ‘not’ may be considered to be not related.”

Updates and Assessing Expectedness

 There is additional guidance about updating the RSI and about determining expectedness for expedited reporting and the DSUR, and I strongly recommend that you visit the links below and read the document for yourself.


CTFG Q&A document:

CTFG Q&A document implementation note:

5 thoughts on “New Requirements for Reference Safety Information in the EU from January 2019

  1. Hi Jo. Thanks for this useful summary. A couple of points …

    Firstly, the second sentence of your point 3.3 appears to be a little ambiguous. I imagine that you probably mean that “You can’t include events JUST BECAUSE THEY might be anticipated from the pharmacology of the product or the product class (IF THEY HAVE NOT BEEN OBSERVED WITH THE ACTUAL PRODUCT)”. As written, it could be taken to mean that one should not include in the RSI events which _had_ been observed with the product if such events could have been anticipated from the pharmacology of the product or product class. Could you clarify?

    Secondly, I have been aware of situations in which an IB has mentioned one or more Serious suspected ADRs which have been observed (in more than one subject) with the product but, because it related to a very early study (i,e, very few exposed subjects) of very novel products, the IB also stated that no events would be regarded as Expected, so that all Serious suspected ADRs would be regarded and treated as SUSARs. How does such an approach sit with what you have said about the upcoming new requirements?


    1. Jo Burmester

      Hi John
      Thanks for your comments. You’re right about point 3.3, you can of course include events which could be anticipated from the pharmacology if they have been observed with the actual product.
      With regard to your second point information about SSARs can be included in the IB without being part of the RSI – this illustrates the importance of having a separate, specific RSI section which is used for the expectedness assessment.

      1. John Whittington 7th January 2019 — 1:33 pm

        Thanks for the confirmation, Jo. As for …

        “With regard to your second point information about SSARs can be included in the IB without being part of the RSI – this illustrates the importance of having a separate, specific RSI section which is used for the expectedness assessment.”

        …I wonder if those who write IBs should perhaps give more thought to their wording in the situation I mentioned. In the examples I recall, the IB says that, because of the nature of the product and the very early status of the study, all SARs will be regarded as “unexpected” (hence the event a SUSAR), even if they have been observed more than once with the product in question (far from impossible, particularly if the event is ‘predictable’ from the pharmacology etc.).

        I assume that (per the ‘rules’), if the event had been observed more than once with the product, it would have to be included in the “RSI section”, and therefore (at least, usually) would be the basis of the expectedness assessment (i.e. the event would be regarded as “expected”, hence the event not a SUSAR). There would therefore be a contradiction between the RSI section and other statements in the IB. In such situations, it is clearly the wish/intention of the sponsors concerned that such events should be subjected to expedited reporting, but I’m not sure how the regulatory authority would feel about an expedited report which related to an event which, in terms of the product’s actual RSI, is not a SUSAR. Can you think of a tidy way of handling that situation (particularly in relation to the writing of the IB and RSI)?

        Kind Regards, John

      2. Jo Burmester

        Hi Again John
        I’m by no means an expert on this, but the ‘rules’ don’t require inclusion in the RSI after an event is observed more than once – that is the minimum requirement for inclusion, but it’s up to the sponsor to decide at what point an event should be included in the RSI and therefore become expected. In the situation you describe the simplest solution is to describe the events in the IB and explain that they are not, at this stage, being classed as expected. Those events would then NOT be included in the RSI section until such time as the decision is made to class them as expected.

      3. John Whittington 7th January 2019 — 4:16 pm

        Fair enough, Jo – I confess that I had not noticed that there was not actually a requirement to include in the RSI events that had been observed more than once with the product

        I must say that this rather changes my view of the RSI, since it seems to imply that the _only_ purpose of the RSI is to define expectedness for the purpose of expedited reporting (and classification in DSURs etc.) – I had previously assumed that, in addition to that purpose, it was also intended as a means of bringing to the attention of investigators SARs which they might ‘expect’ (everyday sense) to possibly encounter with the product. It now seems that an investigator cannot rely on the RSI alone to necessarily provide that information.

        I suppose this is consistent with with your original Point 4, about fatal and life-threatening events. I’m not sure that I necessarily agree with the argument that such events should not appear in the RSI because sponsors should not expect fatal or life-threatening events – since, as you imply, in some therapeutic areas and with some classes of products, such events are not necessarily avoidable, and the risk of them occurring may be ‘accepted’ on the basis of risk-benefit analysis. However,. I can understand a desire for all fatal and life-threatening events to be subjected to, for example, expedited reporting – so if the sole intended purpose of the RSI is to indicate what serious events do NOT require such reporting (rather than also ‘to inform’), one can understand such events being excluded from the RSI.

        In practice, of course, things are more complicated, and certainly less ‘tidy’. Whether or not a particular event is fatal or life-threatening (or, indeed, whether or not it is even Serious) will often/usually depend upon the severity of the event (not to mention other ‘chance’ factors), rather than just its nature. It is therefore very possible that an event would be correctly included in the RSI because it had been reported (more than once) as a SSAR of a severity not great enough to render it fatal or life-threatening. If the (qualitatively) same event was then reported subsequently, with a severity which caused it to be fatal, or for it to be judged life-threatening, would the event then theoretically have to be removed from the RSI? This is not hypothetical – I can think of countless events to which it might apply – haemorrhage, bone marrow suppression, renal/hepatic dysfunction, cardiac problems etc. etc.

        If the sole intention had been to require that all fatal and life-threatening events should be subjected to expedited reporting etc., it would probably have been simpler (better?) to just say so, explicitly, without regard to considerations of ‘expectedness’!

        Kind Regards, John

Leave a Reply

%d bloggers like this:
search previous next tag category expand menu location phone mail time cart zoom edit close